An Integrated Logistics System for Effective Resource Distribution in Post-disaster Humanitarian Relief Operations

Abstract

After a disaster, distributing supplies and transferring people are critical operations and should be done quickly and fairly, with consideration to difficulties associated with limited resources. In this research, more realistic and integrated models are proposed to perform several important logistic operations, including commodity distribution, wounded evacuation, and work–force transfer. To accomplish this, first, an existing model of Yi and Kumar (2007), which considers two logistic operations (e.g., commodity distribution and wounded evacua- tion), is discussed and corrected. Second, a new model is developed to incorporate a new logistic operation of work–force transfer. Evaluation of this model shows that it has some of the same limitations as the Yi and Kumar (2007) model such as lacking detailed vehicles routes. Third, an integrated logistics system is developed to incorporate all three logistic operations while considering realistic issues such as the determination of detailed vehicle routes. Tiny-scale problems are solved optimally via CPLEX-Concert Technology, while problems of realistic size require the application of new heuristic approaches based on solving the model iteratively and optimally according to specific routes which are constructed greedily to achieve the maximum resources utilization. Different heuristic versions are considered to solve the model, the performances of which are compared to the CPLEX results for numerous randomly generated data sets, and they show excellent results in an extremely short processing time. Local search is used in conjunction with replacement and insertion to improve the suggested solution approaches. In replacement, one customer visit could be replaced by two customer visits in the existing routes, if possible, to increase the number of visits which improves the distribution system. Using insertion, a node may be added to existing routes, if possible, to improve the efficiency ii of vehicles. These searches are applied in different ways and the results show that applying them for each candidate solution with higher numbers of iterations (longer termination time) gives the best results among all ways. Finally, a more comprehensive, multi-objective model is developed to consider the use of large vehicles as temporary satellite facilities, serving as mobile supply nodes to improve the efficiency of smaller vehicles. The objectives are considered separately which will minimize the total wounded deviation, the total worker deviations, or the total commodities deviations. Different approaches are developed to find a wide range of solutions for more representative Pareto sets. It is found that there are some clusters in both wounded deviations–worker deviations and wounded deviations–commodities deviations Pareto sets, but they decrease in the commodities deviations–worker deviations Pareto set. Despite the problem of clusters, the suggested solution approaches are capable of finding many solutions in different regions of Pareto sets to cover most cases that might be requested from users.